Dusting myself off, getting back on the horse, doing it again.

[compilerbitch]

As I didn't want to let what happened yesterday get to me (see this if you've not read it already), I've jumped into a new photographic/visual art project. It's one I probably could never show at a camera club, partly because the content is probably going to be too edgy for the octogenarians in the audience, and partly because some of it isn't actually going to be photographs as-such anyway.

Here's the plan, cut & pasted from my wiki:

Screwed

From SarahWiki

The idea behind this project came up whilst thinking about ways to move on from the 'impossible' images I've been making recently (see Sliced for more about that). I think that the thing that got me with those images was the idea of creating images that look superficially real, but that on further examination reveal themselves to be impossible, even though (hopefully) I've managed a sufficiently photorealistic result that it isn't necessarily clear how I managed it.

The Screwed project extends this, but rather than being a single image, I'm intending to make it a more extensive portfolio. Partly this is because I have a lot of ideas and can't really definitely decide upon one single approach, but mostly because I think the idea will stand to become a portfolio of between 6 and 12 images. In the manner of a day-by-day trip report, I'll include a blow by blow account of each image as it's created here, with some separate 'how I did it' pages for the more complicated rigs. Once it's complete, I'll include a 'photos only' version of the page so I can get all high-falutin' fine art about it. As usual, all of the images here are being prepared at print resolution (minimum 4000x3000), so bear in mind that the versions here have about 1/25th that resolution.

Art Concept

I want to use the concept of screws -- wood screws, self-tapping screws, machine screws, etc. -- in contexts that are likely to cause an emotional impact on the viewer. In terms of visual style, this is going to be a B&W project, shot mostly with the Bronica/Megavision system, though with extensive CGI elements from Lightwave. I'm aiming for something somewhat darker than much of my previous work, in concept if not literally. Compositing, as usual, will be in Photoshop CS2. Something that will be a considerable departure for me is that -- gasp -- I'm going to use models. As in human models. With screws. Really, it'll be much cooler than it probably sounds.

Test renderings

At this stage, I'm throwing around visual ideas. I'm a bit torn about whether I should make sets and photograph actual screws, or whether I should just go for CGI straight off. If CGI gives me the look I want, it has a lot of advantages over the real thing -- with CGI, you have a lot more wiggle room in aligning things before you do a final render, so it does have some advantages in compositing. For now, here are a couple of pure CGI concept images to be getting along with:

 

Screwed Perspective

I made a 3D model of a typical cheese head, flat blade self-tapping screw. This was my first test render, just showing a single screw cutting through a flat plane. I did a bit of dramatic lighting, and made a procedural texture for the plane that emphasised the perspective (hence the name). The screw material is basically a fairly standard metal model, raytraced reflections with a spherical reflection map. I'm doing some depth of field simulation -- initially I had the quality turned right up, but I actually prefer the multi-imaged look of the lowest quality option somehow. I like this image -- it might well end up in the final portfolio in its own right. And yes, I did notice that it had a left-handed thread.

 

Mushrooms

This is a radiosity rendering, again with some depth of field simulation, postprocessed to look like it was shot with a 35mm camera with push-processed black and white film. Again, I quite like the image. I'm not sure if this kind of rendering will fit with the rest of the project, but we'll see.

Comments

[shermarama]

Eeep, practical brain kicks in. Some of it looks very real, of course - I think you're right that the low-quality, slightly blocky texture looks better, because it makes the surface a little uneven-looking, like a real cheaply-struck screw. That realness leaves two unreal things standing out. One, there's no distortion of the material around where the screw goes in, not even any slight suggestion of it, which makes the screw look like it's been sliced off and stuck on to the wall - or perhaps that's what you're aiming for. In which case the other unreal thing gives it away, which is the perfect unblemished slot - they're never made dead accurate and by the time you've used it even once there'll be a little distortion, out at the radius of the screwdriver.

With that out of the way, I'll be interested to see what happens with this. What do you use for the 3D stuff?

[arrhythmia.livejournal.com]

great! looking forward to seeing these.

(aside: i immediately thought of this.)

[compilerbitch.livejournal.com]

This was intended to be a quick test render, mostly to check out the screw model, so I didn't bother modeling any distortion to the plate. It would have been easy enough to do. I'll probably do that actually if I decide to keep this image for the final portfolio.

As for the unblemished slot, I'm OK with that -- I actually want a slightly 'realer than real', overperfect feel. It's also why I gave it a left-hand thread -- I want to mess subtly with people's perception.

The 3D models and rendering were all done in Lightwave. It's not as fashionable these days as 3D Studio or Maya, but it has a user interface that is somewhat less psychotic and a (very good) renderer that's rock solid and that doesn't crash all the time. I almost swapped to 3DS a while ago, but when 8 out of 10 attempted renders resulted in Mental Ray exploding horribly I gave up and went back to LW. I'm happy with LW -- I've used it for more than 10 years now on and off.

[fatdog.livejournal.com]

what's the 'concept of screws'?

[compilerbitch.livejournal.com]

I'm not sure. Maybe it's a bit like a murder of crows or something?

[compilerbitch.livejournal.com]

I'll post 'em as I make 'em!

[compilerbitch.livejournal.com]

Actually the surface on the first screw has a procedural bump map on it that looks like hammered metal -- I'm using it at quite a low setting (just about 8%) so it just subtly messes up the surface so you don't get quite clean reflections from it. That's happening in addition to the low-fi depth of field simulation.

I like making CGI images -- you really get to fiddle with things at an incredibly detailed level. I usually have very clear ideas about what I want to see in the final image, so CGI can be a lower pain option than finding a physical thing and photographing it sometimes.

I've had the odd bad reaction to the images I've made. I had one guy (on the Luminous Landscape forum) have a go at me, saying that my work had no artistic merit because I was using CGI. His main reasoning supporting his argument was that the second and third Matrix movies sucked. Umm.

[randwolf.livejournal.com]

I like the concept a lot, dislike the too-perfect form of the screw, but maybe it'll grow me. At the risk of harping on the subject, I think you'll get a much better-looking result using Radiance for the screws; the modeling of metals and global illumination is, I think, superior to either of your studies.

[shermarama]

Aha, yes, the subtly-hammered look definitely works. From the perspective of aiming for art I still don't like that slot as it is, myself - absolutely, leaving some unreal things as clues is a part of what makes it art but whereas the lack of material distortion around where the hole would be gives you two possible interpretations of what you're seeing, the unreal slot doesn't bring anything extra, like it just hasn't been sorted out rather than being an artistic choice. But then this is a random trial render anyway, I realise, it's just what struck me.

I can see your point about CGI, it can get very absorbing...

[technolope.livejournal.com]

Hi! I jumped in here after finding you on amitufteornot. It seems that we are rather similar beings.

Anyways, I have a comment about your observation of the depth-of-field quality. I saw a presentation at SIGGRAPH by Pixar on "bokeh", which they use to characterize real camera blur. It may be that the high quality setting gives perfect bokeh, while what you like is a more realistic, old-camera bokeh.

I don't know the math, but you might want to look into that.

(And I second the Radiance comment below, though it'd take a while to get good enough to recreate the first image.)